[Spacewalk-list] Spacewalk-list Digest, Vol 131, Issue 32

[Paul-Andre Panon]

On Sun, 28 Apr 2019 23:04:36 -0600 , Meena Rajani <meenakrajani at gmail.com>
wrote;

>Hi All:

>I am new to spacewalk. I have  a large cluster of hadoop nodes which I
want to patch using spacewalk. I
>want to patch systems serially. for example if I have 3 systems, Ab,B and
C I would like to stop application
>services on one server A and patch run the application software perform
some checks and then only if the
>system and cluster is in healthy state I would like to go ahead and
repeat the steps for the B and C. My
>question is how do I add my hdfs commands and shellscript before the
patching and after the patching?

>Regards

>Meena

Maybe you should just script the whole thing from the shell using
combinations of spacecmd and hdfs commands instead?

For instance you could build a list of your servers with
group_listsystems, then loop through the system list to perform the
following on each system:
a) approve patches with system_upgradepackage
b) wait and check that system_listupgrades indicates there are no more
packages to be applied
c) trigger a reboot with system_reboot if it's needed
d) wait a couple of  minutes + until hdfs shows the cluster has stabilized

The tricky part is figuring out if the reboot is needed. I don't know if
the situation has changed since
https://www.redhat.com/archives/spacewalk-list/2016-November/msg00012.html
. You might have to just make that an option on your script, or just
always trigger the reboot.

Cheers,

Paul-Andre